[Eeglablist] Epoch length when computing ERSPs

[Klatt, Laura]

Hi all,

I have a question regarding the resulting epoch length when computing ERSPs with the EEGLAB STUDY functions.

According to the formular outlined here (https://sccn.ucsd.edu/mediawiki/images/b/b4/Slide3.jpg, image from https://sccn.ucsd.edu/wiki/Makoto's_preprocessing_pipeline#Why_is_my_ERSP.2FITC_data_shorter_than_ERP_epoch.3F_.2808.2F20.2F2021_added.29), the window width when computing ERSPs should be 1000 ms / min_freq * min_cycles.

I chose the following parameters as input for std_precomp():
[STUDY, ALLEEG] = std_precomp(STUDY, ALLEEG, {},'ersp','on','erspparams',...
    {'cycles', [3 0.5], 'freqs', [4 30], 'nfreqs', 52, 'timesout', 300, ...
    'baseline', [-600 -400], 'basenorm', 'off', 'trialbase', 'full'} ,...
    'itc','on');

Thus, in my case, the formular above yields 1000 / 4 * 3 = 750. Hence, I should "lose" 375 ms on either end of my epochs. With an initial epoch length of [-1000 2000] I should end up with an epoch of -625 to 1625 ms, but after precomputing ERSPs with the EEGLAB STUDY, I end up with epochs that range from -580 to 1580 ms. Does anyone have an idea why?

Additional info: I use EEGLAB v2021.1 and MATLAB R2021b. The input data has a sampling rate of 250 Hz.

I appreciate any hints and comments.

Best,
Laura

--
Dr. Laura-Isabelle Klatt
Leibniz Research Centre for Working Environment and Human Factors
Department of Ergonomics
Research Unit: Information processing
Ardeystr. 67
44139 Dortmund

Contact:
klatt at ifado.de
+49 / 231 1084 260
twitter: @LoraKlatt

Personal website: https://urldefense.proofpoint.com/v2/url?u=https-3A__lauraklatt.github.io_&d=DwIGaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=FmQKobFwPmBUdnSue2E2FVP6Pxzg8V3xIDxr0vUchKBKoGXz8qizEAFK1EyGPOy-&s=QehV0xLaf820LWRQxTKW7ZsxmdhB11dKKruZkLvLOzY&e=